ADVANCED LINER COOLI NG NUMERICAL ANALYSI S FOR LOW ...

25TH

INTERNATIONAL CONGRESS OF THE AERONAUTICAL SCIENCES

1

Abstract

The aim to reach very low emission limits has

recently changed several aspects of combustor fluid

dynamics. Among them, combustor cooling

experienced significant design efforts to obtain

good performances with unfavourable

conditions. This paper deals with simplified 1D

and complete 3D conjugate numerical

simulations of effusion cooling configurations,

performed in the first two years of the European

research project INTELLECT D.M. Geometries

are derived from typical LPP combustor cooling

configurations, which feature low coolant mass

flow rate and high pressure losses (compared to

typical blade cooling parameters).

Results are obtained in terms of local

distributions of effectiveness and discharge

coefficient. Comparison among simulations

allowed to derive useful indications on overall

effectiveness behaviour.

The configuration simulated in this paper

represents a combustor liner with effusion

cooling: the plate tested on the CNRS-LCD test

rig of Poitiers is composed by two different

patterns of effusion cooling. Furthermore, to be

the most representative of a combustor

chamber, an air flow bleed at the exit of the cold

flow is introduced.

On the investigated plate SNECMA

performed 3D conjugate (coupling

fluid/thermal) calculations using a 3D CFD

code named N3S-Natur and ABAQUS, a well

known 3D thermal code. The codes take into

account the effusion cooling area as an

homogenous wall described by a permeability, a

discharge coefficient for the CFD code and a

convective flow (hcon, Tcon) for the thermal one.

That means that such simulations are not

solving the flow inside each hole.

The fluid code also enables to compare the

experimental adiabatic effectiveness

measurements on this plate, but the aim is

before all the overall effectiveness.

Conjugate calculations were also

performed by means of a procedure employing

1D correlative fluid analysis and 2D metal

conduction study. Finally, complete 3D CFD

conjugated calculations has been carried out on

the plate to verify the validity of assumptions

and results obtained with simplified approaches

previously exposed.

1 Introduction

Over the last ten years, there have been

significant technological advances towards the

reduction of emissions, strongly aimed at

meeting the strict legislation requirements.

Some very encouraging results have already

been obtained but the reached solutions have

created other technical problems. Modern

aeroengine combustors, mainly LPP-DLN (Lean

Premixed Prevaporized Dry Low NOx), operate

with premixed flames and very lean mixtures,

i.e. primary zone air amount grows

significantly, while liner cooling air has to be

decreased [1]. Consequently, important

attention must be paid in the appropriate design

of combustor liner cooling system; in addition,

further goals need to be taken into account:

reaction quenching due to cool air sudden

mixing should be avoided, whilst temperature

distribution has to reach the desired levels in

terms of both pattern factor and profile factor

ADVANCED LINER COOLING NUMERICAL ANALYSIS FOR LOW EMISSION COMBUSTORS

A. Andreini *, J.L. Champion **, B. Facchini*, E. Mercier***, M. Surace*

*Department of Energy Engineering “Sergio Stecco” – University of Florence, Italy,

** LCD-CNRS, Poitiers France

*** SNECMA, Villaroche, France

Keywords:liner, cooling, effusion

ANDREINI, CHAMPION, FACCHINI, MERCIER, SURACE

2

[2]. Among various possible techniques to

guarantee an effective liner wall protection,

effusion cooling certainly represents one of the

most promising solutions. Effusion represents

the evolution of classical film cooling solutions

where the scheme is based on a limited number

of injection rows, recently improved by the

introduction of holes with very complex

geometry, at least at the exit. Only in recent

years, has the improvement of drilling capability

allowed to perform a large amount of extremely

small cylindrical holes, whose application is

commonly referred as effusion cooling. Even if

this solution does not guarantee, for each hole,

the excellent wall protection achievable with

film cooling, the most interesting aspect is the

significant effect of wall cooling due to the heat

removed by the passage of coolant inside the

holes [3]. In fact, a higher number of small

holes, uniformly distributed over the whole

surface, permits a significant improvement in

lowering wall temperature. From this point of

view, effusion can be seen as an approximation

of transpiration cooling by porous walls, with a

slight decrease in performance but without the

same structural disadvantages. Particularly for

combustor liners, such solution appears very

interesting because radiation contributes

significantly to the heat flux, not sufficiently

reduced by the film cooling alone. Studies on

effusion cooling, or on multi-row hole injection

have been performed to understand the complex

phenomena which the effusion is based on, and

experimental analysis appears fundamental [4,

5, 6, 7, 8, 9, 10]. The use of CFD is very

complex for film cooling and particularly for

effusion, because standard RANS (Reynolds

Averaged Navier-Stokes Simulation) approach,

with common turbulence models, is not able to

reproduce correctly film effectiveness

distribution, so more advances approaches as

DES (Detached Eddy Simulation) or LES

(Large Eddy Simulation) to solve Navier-Stokes

equations are necessary [11, 12, 13, 14]. In the

effusion cooling analysis, the conjugate

approach to solve simultaneously heat

convection and conduction appears very useful

to better understand phenomena and to estimate

cooling performances [15, 16, 17]. From the

design point of view, the effusion requires a

simulation tool in order to predict overall

effectiveness, whenever boundary conditions

and geometry parameters change. This could

permit to properly design the hole array

geometry depending on the location and the hot

gas thermal loads. Such a tool can be developed

basing on well known correlations about heat

transfer inside the holes to predict film cooling

adiabatic effectiveness [18, 19, 20, 21, 22]. The

aim of this paper is to study a liner effusion

cooling geometry comparing experimental

analysis with CFD conjugate analysis, including

some tests with the 1-D design tool. The

objective of INTELLECT D.M. project is to

develop a design methodology for lean burn low

emission combustors to achieve a sufficient

operability over the entire range of operating

conditions whilst maintaining low NOx emission

capability. A specific work package inside the

project is dedicated to the study of advanced

liner cooling systems.

NOMENCLATURE

C = CO2 Concentration

Cd = Effusion hole discharge coefficient

DP = Pressure drop [Pa]

h = Average heat transfer coefficient [W m-2

K-1

]

k = Turbulent kinetic energy [m2/s

2]

M = Mach number

Re = Reynolds number

p = Pressure [Pa]

Q = Mass flowrate [kg/s]

T = Temperature [K]

x = Streamwise distance [m]

y = Distance for the plate [m]

Greeks

ε = Turbulent dissipation [m2/s

3]

φ = Heat flux [W/m2]

η = Effectiveness

µ = Viscosity [kgm-1

s-1]

ρ = Density [kg/m3]

Subscripts

1st = Grid first cell

1,2,3 = Section number

ad = Adiabatic

aw = Adiabatic wall

c = Coolant

con = Convective

g = Gas

i = Inlet, Input

3


h = referred to the holes

loc = local

ov = Overall

x = Abscissa in the streamwise direction

t = Turbulent

wall = Hot side wall

Superscripts

+ = normalized

2 Experimental analysis

2.1 Test rig

The experimental study is performed

using the THALIE test rig [23], whose

schematic view is given in figure 1.

Q 1

Q 2

Q 3

Figure 1: Schematic view of the experimental set up and

of the THALIE’s test section

THALIE allows aero-thermodynamical

conditions close to those encountered in a

combustor (T1 up to 1200 K, T2 up to 600 K).

Primary burnt gases are generated by a tubular

kerosene combustor and the secondary flow

(coolant) is heated by an electrical heater.

However, the present work has been carried out

with T2 = 273 K.

Main flow: subscript “1”

Cross section 123×72 mm²

Reynolds number 95000 < Re1 < 272000

Cooling flow: subscript “2”

Cross section 100×20 mm²

Reynolds number 5300 < Re2 < 108000

Hole flow: subscript “h”

Reynolds number 0 < Reh < 15000

Pressure ratio Dp/p < 5%

Mach number Mh < 0.2

Table 1: Subscript 1 refers to the hot primary flow while

subscript 2 refers to the secondary flow of cooling air

Available diagnostics are: Laser Doppler

Velocimetry, optical measurements (Planar

Laser Induced Fluorescence, Infra-Red

Pyrometry), thin thermocouples, local gas

analysis and flow visualizations. This

metrology allows the determination of velocity,

concentration and temperature inside the wall

film as well as the wall temperature.

As shown in Figure 1, the combustor wall

sample is placed in the rectangular test section,

in such a way that hot burned gases (primary

flow) and cooling air (secondary flow) are

flowing parallel on each side of the solid

separation. The variation ranges of the flow

characteristics are given in table 1. Each of

these parameters can be fixed independently and

is monitored via PID controllers. This ensures

reliability and reproducibility of the

experimental conditions. Moreover,

displacements of measurement devices and data

acquisition are computer monitored. P1(Pa) 3 bar

T1 (K) 1000 K

Q1 (kg/s) 0.3

Re1 56000

Q2 (g/s) 0.1

T2 (K) 273 K

Re2 min / max 56000-75000

Table 2: Experimental conditions

2.2 Experimental conditions and procedures

Experiments have been carried out with

T1 close to 1000 K and T2 close to 273 K The

Reynolds number relative to the primary hot

flow (Re1) is fixed to 56000, while several

values of Re2 have been investigated. By

changing the pressure loss step by step or

continuously, the mass flow rate through the

wall has been changed, allowing the


4

determination of Cd versus Reh at fixed values

of Re1 and Re2.

Experimental conditions are detailed in

table 2. During all these experiments, the

pressure loss remains lower than 5% to avoid

compressibility effects inside holes. The Mach

number (M1) of the primary flow remains lower

than 0.2. Mass flow rates (Q1 and Q2) are

controlled by using PID controllers ensuring

reliability and stability of the experimental

conditions. Mass flow rates (Q1, Q2 and Q3) are

measured by Vortex flow-meters, the

inaccuracy is less than 1%. Flow temperatures

(T1 and T2) are measured by K-type

thermocouples in the inlet section of the

corresponding channels. The temperature T2

results from the air expansion from the storage

pressure (between 100 to 180 bar) to the

atmospheric value. It is therefore subject to

small changes during experiments. The static

pressure (p1) is measured by a transducer while

the static pressure difference (Dp) is given by a

differential transducer in the range 0-500 hPa.

All these data (Q1, Q2, Q3, T1, T2, P1 and Dp)

are acquired at the rate of 1 Hz and stored on a

computer for post processing. Flow parameters

(Re1, Re2, Reh) are calculated using average

values of measurements.

2 different patterns (2 mm thick)

Pattern 1 (D = 0.6 mm)

(X/D)=6.5 ; (Y/D)=6.6

Pattern 2 (D = 0.4 mm)

(X/D)=9.25 ; (Y/D)=10

Figure 2: Experimental geometry

CO2 concentration profiles at different

streamwise locations have been performed

normally to the wall by gas sampling. The

knowledge of the wall CO2 concentration (Cw)

and the maximum CO2 concentration measured

inside the primary hot flow allows to calculate

the non-dimensional wall CO2 concentration

( )1 1* locC C C C= − . According to this latter

expression, C* reaches its minimum value as

the CO2 concentration reaches its maximum.

Finally, it must be noted that the value of C* at

the wall gives the adiabatic cooling

effectiveness ( )1 1wallη C C Cad

= − . The

corresponding error on ηad attached to this

determination is found to remain lower than 5%.

Geometry is presented in figure 2.

3 Numerical calculations

3.1 SNECMA methodology

Considering the configuration of cooling

system described before, CFD calculations have

been performed to cross-check the experimental

measurements of adiabatic effectiveness. The

3D CFD code used is N3S-Natur.

Figure 3: detail of boundary layer mesh

3.1.1 Calculation meshes

Two kinds of meshes have been realized

for those calculations. The first one only takes

into account the hot flow: the boundary

condition of mass flow rate through the effusion

cooling system is described thanks to an in-

house 1D code.

The second mesh takes into account both

flows. The CFD code, using a porosity

condition, is able to calculate the mass flow rate

through the effusion cooling system.

In order to have a good description of the

wall laws, the meshes have been realized taking

5


into account a kind of boundary layer: the first

layer is 0.2mm high (figure 3).

In figure 4 a complete view of both meshes

is shown.

Figure 4: Overall view of calculation meshes used in

SNECMA methodology

3.1.2 Boundary conditions

As previously said, the hot flow mesh need

an in-house 1D code named GECOPE to impose

the right mass flow rate through the effusion

cooling system. Indeed, those two kinds of

effusion cooling patterns have not the same

description (holes diameter, distance between

two rows especially). To couple the

aerodynamic calculation with the thermal wall

one, the boundary condition of effusion cooling

also included a wall law.

The CFD code N3S-Natur needs different

data to take into account the description of the

effusion cooling:

• mass flow rate (only for the hot flow mesh)

• temperature

• permeability: holes area / total area

• discharge coefficient

• holes inclination

• turbulence (k, ε) (only for the hot flow

mesh)

For all boundary conditions the turbulence

rate has been taken equal to 5% and

200t

µ µ = . That leads to k=2.5 and ε =70 for

the hot entrance, and k=3.8 and ε=155 for the

cold one.

3.1.3 Adiabatic effectiveness calculations

Figure 5: passive effluent field

Figure 6: temperature field

Hot flow mesh calculations

Results are reported in terms of adiabatic

effectiveness defined as:

aw g

aw

c g

T T

T Tη

−=

− (1)

The adiabatic effectiveness is modeled

using a passive effluent introduced for each

effusion cooling boundary condition. We can

see that the cold flow is well introduced and


6

enables to cool the air near the wall. Figures 5

and 6 show simulation results in terms of

passive effluent and temperature fields.

Both flows calculations

The passive effluent that enables to follow

the mixing of both flows through the effusion

cooling is introduced at the cold entrance.

Indeed, this condition calculates the mass flow

rate through the effusion cooling and the

mixing.

The global mass flow rate calculated is 37.1 g/s to

compare with 39.6 g/s for the experimental

measurements.

adiabatic effectiveness at different highs (porosity condition)

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14

x(m)

eta

z= 0.3 mm z=0.5 mm z=0.7 mm z= 1 mm measurements z=0.4 mm

Figure 7: adiabatic effectiveness along the wall for

porosity calculation

adiabatic effectiveness at different highs (hot flow mesh)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

-0.01 0.01 0.03 0.05 0.07 0.09 0.11 0.13 0.15x(m)

eta

z=0.3 mm z=0.5 mm z=0.7 mm z=1 mm measurements z=0.4 mm

Figure 8: adiabatic effectiveness along the wall for hot

flow mesh

The experimental measurements of the

adiabatic effectiveness make use of a 0.5mm

high probe that aspires the gas. To compare with

the 3D calculations, we should post treat the

calculations at different distances from the wall.

Figures 7 and 8 show the ηad distribution for

both the calculation approach in comparison

with experimental data.

The first measurement just after the second

row of holes overestimates the cooling level

because the sample is taken just behind an hole

whereas the film cooling is not well developed.

The optimal distance from the wall is

0.3mm for both calculation.

3.1.4 Thermal calculations

Thanks to convective hot boundary

conditions, a thermal calculation, using the code

ABAQUS, has been performed. The great

interaction between the CFD code and the

thermal one is due to the formulation of the

convective flux. Thus, iterative calculations

have been performed to take into account this

coupling.

Extraction of convective variables for the first

iteration

The first calculation is adiabatic: 0wallφ = .

hcon is the convective coefficient, Tcon the

convective temperature and Taw the adiabatic

wall temperature. So:

0 ( ) 0wall con aw con con awh T T T Tφ = ⇒ − = ⇒ = (2)

So the convective temperature is equal to

the adiabatic wall temperature.

A second calculation is done, by imposing

a wall temperature of Tcon + 100K.

co

( )

( 100 ) 100

wall con w con wallcon

con con n

h T Th

h T T

φ φ= − ⇒ =

= + −

(3)

So the couple (hcon, Tcon) is extracted from those

2 calculations.

Extraction of convective variables for others

iterations

The same methodology could be used for

others iterations:

2 1

1

12

2 1

( ) 0

( ) 0

concon wi con

con wi concon wi

hh T T X

Xh T X TT T

φ φ

φ

φφ

φ φ

−== − =

⇒ = + − = = −

−

(4)

Using directly the wall laws, we can extract the

convective variables :

7


1con st

p p f

con

T Tc u

hT

ρ+

=⇒

=

(5)

In figure 9 a temperature contour map for

the last iteration is reported, while figures 10-12

describe the convergence of the overall

methodology.

Figure 9: Hot side wall temperature at last iteration

convective temperature for each iteration

250

350

450

550

650

750

850

950

1050

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99X (mm)

Tco

nv(K

)

Tconv_ite1

Tconv_ite2

Tconv_ite3

Tconv_ite4

Figure 10: Convective temperature

We can notice that the convergence of the

convective variables are fast and easy. We can

consider that the third iteration is already

converged.

3.2 Correlative approach

3.2.1 Description

In order to evaluate the accuracy of

literature correlations employed in the

preliminary design of film/effusion cooling

systems, a correlative 1D procedure, coupled

with a 2D FEM thermal solver, was set up.

convective coefficient for each iteration

200

300

400

500

600

700

800

900

1000

1100

1200

1 8 15 22 29 36 43 50 57 64 71 78 85 92 99

X (mm)

Hc

on

v

Hconv_ite1

Hconv_ite2

Hconv_ite3

Hconv_ite4

Figure 11: Convective coefficient

hot side wall temperature for each iteration

300

350

400

450

500

550

600

650

700

750

1 10 19 28 37 46 55 64 73 82 91 100X (mm)

Tw

all

(K

)

Twall_ite1

Twall_ite2

Twall_ite3

Twall_ite4

Figure 12: Hot side wall temperature

Such methodology allows to quickly

evaluate the metal temperature distribution of a

generic effusion cooled plate, so as to

investigate wide variations of design parameters

(hole spacing, angle, diameter) and boundary

conditions. The procedure, which is described in

details in [18,19], uses ANSYS™ as 2D FEM

solver for thermal conduction within the flat

plate. Hot and cold side boundary conditions

(heat transfer coefficients and adiabatic wall

temperatures) are obtained with standard fully

turbulent smooth pipe correlations. Boundary

conditions of effusion holes are evaluated by

solving a 1D fluid network solver (SRBC code)

which reproduces the main geometric features

of actual geometry. Coolant is considered as a

perfect gas subjected to wall friction and heat


8

transfer, which are evaluated with specific

correlations: the flow field is solved in subsonic

regime starting from boundary conditions

specified at all inlets and outlets in terms of

pressures or mass flow rates, depending on

design specifications. Film cooling effectiveness

on the hot side of flat plate is also calculated by

SRBC code: in this case L’Ecuyer and

Soechting [20] correlation was used to perform

adiabatic film cooling effectiveness, while rows

Seller’s superposition, presented in

Lakshiminarayana [24], is used. The overall

interaction between SRBC code and ANSYS™,

explaining the iterative procedure employed, is

depicted in figure 13.

Figure 13: Flow diagram of 2D correlative procedure

Figure 14 reports a schematic description

of the fluid network used for the analysis of the

effusion cooled plate considered in this work.

With or without imposing Cd

Effusion Holes (1 model for each row)

Inlet duct Outlet duct

Figure 14: 1D fluid network

3.2.2 Results

The main source of uncertainty in this

analysis is the evaluation of pressure losses due

to effusion holes discharge and the estimation of

heat transfer within the holes, which is

responsible for the heat sink effect. In order to

evaluate the impact on overall accuracy of these

two matters, several simulations were

performed with different assumptions:

1. Pressure losses in holes due to friction factor only

2. Pressure losses in holes due to friction factor plus

an imposed discharge coefficient (Cd=0.9)

3. As in point 3 but using an heat transfer

correlation specific for not fully developed pipes

4. Heat Sink effect neglected

5. Film cooling effect neglected (adiabatic

effectiveness set to zero)

6. Pressure losses in holes estimated by imposing

only a discharge coefficient, without wall friction

(Cd=0.73)[21,22]

Flow boundary conditions were imposed as in

experimental tests resulting in effusion cooling

characteristics reported in table 3.

Effusion mass flow rate 0.937 kg/s

Discharge mass flow rate 0.06 kg/s

Blowing ratio 7.3 -

Velocity ratio 2.1 -

Table 3: Flow characteristic calculated by the fluid

network model

Results are reported in terms of overall

effectiveness, defined as:

_

W Gasov

Cool in Gas

T T

T Tη

−=

−

Figure 15 depicts the distributions of

overall effectiveness obtained with the six

hypothesis above described.

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18

1

2

3

4

5

6

Abscissa [m]

No Heat Sink

No Film Cooling

ovη

Figure 15 Overall Effectiveness distributions

9


First of all, it’s important to point out the

reduced influence of the criteria adopted for the

analysis of pressure losses and heat transfer

within the holes. Moreover, it’s interesting to

highlight the dominant contribution of heat sink

effect in the first part of the plate, where film

coverage is partial, with respect to the last part

of the plate where the effect of reduction of the

hot side adiabatic wall temperature prevails.

300

350

400

450

500

550

600

650

700

750

0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18

abscissa [m]

Tw

all

[K

]

SRBC-ANSYS

SNECMA

Figure 16 Hot side wall temperature vs. x direction

Figure 16 reports a direct comparison

between hot side wall temperature predicted by

SNECMA methodology and the correlative

procedure above described. An overall good

agreement is revealed especially in the last part

of the plate where film cooling coverage

dominates. The discrepancy in the first part of

the plate, where heat sink effect prevails, is

probably due to a different modeling of effusion

holes: the distribution of heat sink effect

adopted in SNECMA runs, probably neglect a

smoothing effect to wall temperature which

results in a steeper variation with respect to the

correlative procedure.

3.3 Full 3D CFD conjugate analysis

As a final step in the analysis of the

effusion cooled plate considered in this work, a

full 3D CFD study was performed in order to

obtain a detailed visualization of flow and

temperature fields in the regions where the

interaction between hot gas and coolant jets

takes place.

3.3.1 Calculation tools and models

Calculations were performed using the

industrial CFD code STAR-CD™, v. 3.26

developed and distributed by CD-Adapco.

STAR is a finite volume unstructured solver

with multi-physics capability (coupled solution

of NS fluid equations and Fourier conductive

equations). In this work fluid domain was

solved using a compressible SIMPLE like

algorithm, while flux discretization follows the

Monotonic Advection and Reconstruction

Scheme (MARS) for all the equations with the

exception of continuity equation, where Central

Differences were used. In order to assure a

proper accuracy but keeping acceptable

computational costs, a two equation k-ε model

was selected, using a Two-Layer formulation

for the near wall region (Norris and Reynolds

scheme). Such approach requires about 15 cells

within the boundary layer with a normalized

distance for the first cell (y+) near unity for a

proper discretization of the near wall zone,.

Calculation mesh was realized with an

automatic multiblock structured approach,

implemented in the STAR-CD preprocessor tool

pro-STAR [17]. In order to keep mesh size

below 4 millions elements (which

approximately represents hardware limit) only

the first eight rows of holes were considered.

Figure 17 reports some details of computational

mesh, while figure 18 describes the

computational domain and its boundary

conditions.

Figure 17: Details of calculation mesh (total 3.7 millions

elements


10

Figure 18: Computational domain and main boundary

conditions

3.3.2 Results

Figure 19 compares adiabatic effectiveness

results obtained by 3D CFD calculation and

SRBC code with experimental data.

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

0 0.02 0.04 0.06 0.08 0.1 0.12

abscissa [m]

ad

iab

ati

c e

ffe

cti

ve

ne

ss

[-]

CFD 3D

exp

SRBC-ANSYS

Figure 19: Axial distribution of laterally averaged

adiabatic effectiveness vs. x direction

Spatial coordinate is limited to the range

investigated with 3D CFD. Despite the known

deficiencies of 2 equation turbulence models in

the prediction of the spreading of round jets in

cross flows [17], averaged effectiveness

predicted by CFD calculations fairly agrees with

correlation. On the other hand, experimental

data are slightly under-predicted, but we have to

consider that they are local values, sampled at

different lateral positions.

ηad

[-]

Figure 20: contour map of adiabatic effectiveness

A detailed contour map of adiabatic

effectiveness is plotted in figure 20. Even if

standard k-ε model tends to under estimate the

spreading of round jets, it’s important to point

out how actual effectiveness distribution shows

a visible non uniformity in lateral direction,

which cannot be quantified with SNECMA

methodology or correlative procedures.

Finally, some discussion about conjugate

analysis. In order to keep a realistic heat flux as

boundary condition on the metal side

downstream of the last hole and upstream of the

first one (see figure 21), the average temperature

predicted with SRBC-ANSYS™ procedure was

imposed.

Fixedtemperature condition

8th row

Figure 21: Detail of fixed temperature condition imposed

at outlet metal side

Figure 22 compares overall effectiveness

calculated with full 3D CFD and correlative

procedure.

11


0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.02 0.04 0.06 0.08 0.1 0.12

abscissa [m]

overa

ll e

ffecti

ve

ness [

-]

SRBC-ANSYS

CFD 3D

Figure 22: Overall effectiveness comparison

A good agreement is revealed confirming

the validity of the assumptions. Figure 23

reports a detail of temperature field near a hole:

the complex distribution depicted points out the

importance of a 3D analysis during the design.

T[K]

Figure 23 Temperature contours close the effusion hole

predicted by conjugate analysis

4 Conclusions

The present study has investigated the

cooling performance for a specific effusion

configuration applied to a combustion chamber

liner of a gas turbine aero-engine.

The analysis has focused on the

comparison between experimental results and

three different simulation approaches.

One correlative analysis based on 1-D/2-D

model and two different solutions for a 3-D

CFD approach have been compared.

The experimental adiabatic effectiveness

distribution is fairly well predicted by all the

numerical approaches as well as the coolant

mass flowrate and discharge coefficient of

perforated plate.

The overall effectiveness distribution has

been calculated only by the three mentioned

numerical approaches; only few limited

discrepancies are noticeable, generally due to

different hypotheses for the micro-holes

modeling.

To sum up, the study shows the potentiality

and the limits of the possible solutions for the

effusion cooling numerical approach.

Acknowledgements

The present work was supported by the

European Commission as part of FP6 STREP

INTELELCT D.M. research program, which is

gratefully acknowledged together with

consortium partners.

References

[1] A. Schulz, Combustor liner cooling technology in

scope of reduced pollutant formation and rising

thermalefficiencies, Heat transfer in Gas Turbine Systems,

934, (2001), 135–146.)

[2] A. H. Lefebvre, Gas Turbine Combustion, Taylor &

Francis (1998))

[3] K. M. B. Gustafsson. Experimental studies of effusion

cooling. Chalmers University of technology, Department

of Thermo and Fluid Dynamics,

(www.tfd.chalmers.se/˜lada/postscript_files/bernhard_phd

.pdf ), 2001.

[4] Goldstein, R. J., Eckert, E. R. G., and Ramsey, J. W.,

1968 Film cooling with injection through holes: adiabatic

wall temperatures downstream of a circular hole J. of

Engineering for Power, 90 , pp. 384.395.

[5] Afejuku, W. O., Hayi, N., and Lampard, D., 1979. The

film cooling effectiveness of double rows of holes, ASME

paper (79-GT/Isr-10) .

[6] M. Sasaki, K. Takahara, T. Kumagai, and M. Hamano.

Film cooling effectiveness for injection from multirow

holes Journal of Engineering for Power, 101, 1979.

[7] R. E. Mayle and F. J. Camarata. Multihole cooling

film effectiveness and heat transfer ASME Journal of heat

transfer, 97, 1975.

[8] G. E. Andrews, A. A. Asere, M. C. Mkpadi, and A.

Tirmahi. Transpiration cooling - contribution of film

cooling to the overall cooling effectiveness International

Journal of Turbo and Jet Engines, 3, 1986.

[9] G. E. Andrews, A. A. Asere, M. L. Gupta, and M. C.

Mkpadi. Effusion cooling: the influence of number of hole

Journal of Power and Energy, 204, 1990.


12

[10] M. Martiny, A. Schulz, and S. Witting. Full coverage

film cooling investigations: adiabatic wall temperature

and flow visualization ASME Winter Annual Heat

Transfer, (95- WA/HT-4), 1995.

[11] J. E. Mayhew, J. W. Baughn, and A. R. Byerley. The

effect of freestream turbulence on film cooling adiabatic

effectiveness ASME Turbo Expo, (GT-2002-30172),

2002.

[12] Leylek, J. H., and Zerkle, R. D., 1993. Discrete-jet

film cooling: A comparison of computational results with

experiments ASME Turbo Expo (93-GT-207) 2003.

[13] Roy, S., Kapadia, S., and Heidmann, J. D., 2003.

Film cooling analysis using des turbulence model ASME

Turbo Expo (GT-2003-38140), 2003.

[14] Acharya, S., and Tyagi, M., 2003, Large eddy

simulation of film cooling flow from an inclined

cylindrical jet ASME Turbo Expo (GT-2003-38633).

[15] Bohn, D., Ren, J., Kusterer, K., 2003, Conjugate heat

transfer analysis for film cooling configurations with

different hole geometries ASME Turbo Expo (GT2003-

38369)

[16] Papanicolaou, E., Giebert, D., Koch, R., and Schulz,

A., A conservation-based discretization approach for

conjugate heat transfer calculations in hot-gas ducting

turbomachinery components. International Journal of

Heat and Mass Transfer, 44 , pp. 3413.3429, 2001.

[17] A. Andreini, C. Carcasci, S.Gori, M. Surace, 2005,

Film cooling system numerical design: adiabatic and

conjugate analysis ASME Heat Transfer Conference

Expo (HT2005-72042)

[18] C. Carcasci, B. Facchini, G. Ferrara. A rotor blade

cooling design method for heavy duty gas turbine

applications ASME Cogen-Turbo Power, (95-CTP-90),

1995.

[19] C. Carcasci and B. Facchini. A numerical procedure

to design internal cooling of gas turbine stator blades.

Revue Generale de Thermique, 35, 1996.

[20] M. R. L’Ecuyer and F. O. Soechting. A model for

correlating flat plate film cooling effectiveness for rows of

round holes. AGARD Heat Transfer and Cooling in Gas

Turbines (SEE N86-29823 21-07). September 1985.

[21] M. Gritsch, A. Schulz, and S. Wittig. Method for

correlating discharge coefficients of film cooling holes

AIAA Journal, 36, 1998.

[22] M. Gritsch, A. Schulz, and S. Wittig. Discharge

coefficient measurements of film cooling holes with

expanded exits ASME Journal of Turbomachinery, 120,

2001.

[23] Rouvreau, S., Etude experimentale de la structure

moyenne et instantane e d'un film produit par une zone

multiperforee sur une paroi plane. Application au

refroidissement des chambres de combustion de moteurs

aeronautiques PhD Thesis, CNRS, Poitiers, France 2001.

[24] B. Lakshminarayana,. Fluid Dynamics and Heat

Transfer of Turbomachinery. John Wiley & Sons. New

York, 1996.

ADVANCED LINER COOLI NG NUMERICAL ANALYSI S FOR LOW ...

Documents

Transcript of ADVANCED LINER COOLI NG NUMERICAL ANALYSI S FOR LOW ...